EMRFD Message Archive 5292
Message Date From Subject 5292 2010-09-29 02:18:32 davidpnewkirk Getting useful harmonic output from a BJT Colpitts oscillator I've been modeling (with the Serenade 8.5 SV RF circuit simulator by Ansoft) the July 2006 version of the BJT crystal oscillator described by Wes Hayward in the Micromountaineer Revisited transceiver (July 2000 QST), the Universal QRP Transmitter (April 2006 QST) and "Crystal Oscillator Experiments," a Feedback item on the Universal QRP transmitter (July 2006 QST). The circuit is a common-anode Colpitts design with a parallel-tuned circuit between the BJT collector and ac ground, with output taken from the collector tank by a link.
Wes revisited this oscillator in July 2006 QST because some builders of the Universal QRP design as it was published in April 2006 QST had reported difficulty with not-controlled-by-the-crystal oscillation at 80 meters. Wes found that this mode could be disallowed by reproportioning the voltage-divider capacitors to 270 pF/270 pF and adding a 33-ohm resistor in what had been a direct connection between the BJT emitter and the junction between the voltage-divider caps. This revised version of the Micromountaineer II/Universal QRP oscillator is the version I modeled.
One suggestion I can make about this topology in general is that oscillation modes other than those attributable to the crystal fundamental and nearby secondary resonances aren't that surprising when we consider that we're obtaining output by connecting a high-Z LC network in a lead that "should" be directly ac-grounded (the BJT collector). Especially considering the BJT's changing gain with frequency and the excellent high-frequency response of modern BJTs in general, inserting a parallel-tuned tank where we would expect an ac ground could be expected to provide surprises, the first of which being that the circuit oscillates at the crystal fundamental at all.
But oscillate at the crystal fundamental it does, and if we can grab enough output from the collector, we're ahead in usability (one side of the crystal can then be grounded, allowing easy crystal switching or insertion of L and/or C for frequency pulling) and in efficiency (obtaining the same net output by coupling out of the emitter would likely require an additional amplifier stage).
I waded in with modeling on this circuit because of work I'd done in evaluating a harmonic crystal oscillator described by Frank Jones, W6AJF, in the 1930s. As I suspect many experimenters do after first encountering the Pierce topology, Jones went a little wild with applying the Pierce in almost every transmitter design he originated for a year or three. For vacuum-tube enthusiasts, at least, it's a party that invariably ends with the achievement of the understanding that the Pierce cannot develop appreciable power without greatly stressing the crystal, especially at 7 MHz and above, and that attempting to tune its anode--never mind that doing so throws away the tankless "free lunch" aspect of the circuit--runs us right into stopping its oscillation because anode resonance at the crystal fundamental disallows achievement of the phase shift necessary for oscillation. For a popular application of a boosted Pierce ( http://mysite.verizon.net/dpnewkirk/ej/boosted_pierce/ ) a bit more recent than Jones's 1930s designs, look no farther than Doug DeMaw's Tuna Tin 2.
To me, Jones's coolest contribution to the Pierce-as-master-oscillator canon was his "sure-fire" circuit of April 1938 *Radio*: a triode or tetrode Pierce oscillator with a second-harmonic parallel tank inserted between the tube anode and the crystal, all text descriptions of which, in *Radio* and in the *Radio Handbook*, very studiously avoid referring to its underlying topology as the Pierce. With the version I built with a new-old-stock Taylor T-21 tetrode, I obtained over 5 W at 7 MHz without stressing my 3.5-MHz crystal--albeit an FT-171B-cased job intended for use in a BC-610 transmitter--at all, and with excellent keying.
So it occurred to me that if we can grab useful fundamental output from a common-anode BJT oscillagtor by inserting a parallel-tuned tank in its anode-ground line, we should be able to grab useful *harmonic* output as well. And so here are my modeling results based on Serenade 8.5 SV's BFR193 BJT model, the RLC equivalent of a 3.5-MHz crystal, and a supply volage of 12.
3.5-MHz collector tank (6.6 uH
5294 2010-09-29 06:13:45 Tim Re: Getting useful harmonic output from a BJT Colpitts oscillator There have been a couple of philosophical-and-practical articles about nonlinearities in oscillators in QEX recently.
If you measure current through the crystal (e.g. a transformer or resistor in series with the crystal) you can end up with a beautiful nondistorted (small harmonic content) signal there, while elsewhere in the circuit you have substantial harmonic content. I don't think this is shown too much in EMRFD but it shows up prominently in recent ARRL Handbooks. In general it's not hard to get a lot of harmonic content out of the *right point* in a crystal oscillator. For others the *right point* is the one with low harmonic content :-)
The tube oscillator with harmonic-output designs are wonders of low active parts count (the ECO is just plain brilliant) but with Micromountaineer class rigs I think that power efficiency and predictable operation over Vcc changes might be more relevant than number of active devices.
In a small portable transmitter, if you want controlled even harmonic generation with power gain in the same stage, IMHO it's hard to beat the push-push doubler for overall efficiency. Maybe you have some limit
5295 2010-09-29 08:09:53 William Carver Re: Getting useful harmonic output from a BJT Colpitts oscillator > 7-MHz collector tank (3.3 uH 5296 2010-09-29 14:09:04 joop_l Re: Getting useful harmonic output from a BJT Colpitts oscillator I have done some experiments with a piece of the circuit described here:
Just the part around T1 and its output bandfilter was built and examined. It is also a Colpitts oscillator, albeit a 3rd overtone. The bias level can be adjusted with the potentiometer. It determines how hard the circuit oscillates. As you are probably aware the collector is conducting only part of the cycle. How much is determined by the bias level. Thus it also determines how much of the desired harmonic is present. With a SA connected it is easy to tune the circuit for optimum operation at the 2nd harmonic (here 144MHz).
Joop - pe1cqp
5297 2010-09-29 17:47:50 davidpnewkirk Re: Getting useful harmonic output from a BJT Colpitts oscillator 5298 2010-09-29 18:22:31 KK7B Re: Getting useful harmonic output from a BJT Colpitts oscillator Hi Dave,
This is great stuff. I particularly like your approach of taking a bit of history, thinking it through, making some measurements at the bench, and then applying it to a current problem.
The self-buffering effect is profound when you take the output of an oscillator on harmonics--particularly even harmonics. I think that may have been so well known by the time I was born that authors forgot to menti
5300 2010-09-30 04:32:38 davidpnewkirk Re: Getting useful harmonic output from a BJT Colpitts oscillator 5301 2010-09-30 05:38:28 joop_l Re: Getting useful harmonic output from a BJT Colpitts oscillator Yes, harmonic selection can be done by filtering (the coupled resonator part) as well as the generation (the bias level).
To get some insight in the importance of how to generate the right harmonic, this document might be useful:
Especially look at the graph on page 2.
This might be useful for standalone frequency multipliers as well. Up to now I have not dealt with multipliers much. I tried some stuff with SPICE, but it never seemed really good. Now I know the right signal level is important for best performance in harmonic of interest.
The subharmonic/fundamental impedance of the filter in the collector path should be low enough to allow proper oscillation. So in my circuit I tried to keep the inductor impedance at a few hundred Ohm. The tank capacitor followed from that.
And I also tapped the collector side at a 2nd or 3rd winding whereas the output is connected at the 1st winding. This came from the thought that the collector is of much higher impedance than 50 Ohm output load. It give better matching and as a result more output.
5303 2010-09-30 07:19:06 Bob Re: Getting useful harmonic output from a BJT Colpitts oscillator Hi all -
I've been reading along and have noted the word "subharmonic" several times. It's not clear to me where/how "subharmonics" can originate. Would this not require something like every second, third, .. Nth waveform cycle to somehow be different from the bunch? Or is "subharmonic" always 1/2 Fo? Could such difference be both an amplitude and frequency one? Or? Any thoughts?
5304 2010-09-30 09:18:00 joop_l Re: Getting useful harmonic output from a BJT Colpitts oscillator "subharmonic" is probably a confusion term. Wikipedia refers to the term as frequencies below the fundamental. In this context I read it as anything below the desired harmonic, inluding the fundamental frequency. But the term it is probably best avoided.
5305 2010-09-30 09:33:35 ehydra Re: Getting useful harmonic output from a BJT Colpitts oscillator I think it is not a problem at all:
m stands for harmonics
n stand for subharmonics
Yes, there are subharmonics possible of the harmonics. That is what we
see in reality.
> "subharmonic" is probably a confusion term. Wikipedia refers to the term as frequencies below the fundamental. In this context I read it as anything below the desired harmonic, inluding the fundamental frequency. But the term it is probably best avoided.
5310 2010-09-30 12:44:46 davidpnewkirk Re: Getting useful harmonic output from a BJT Colpitts oscillator 5313 2010-09-30 13:56:40 Bob Re: Getting useful harmonic output from a BJT Colpitts oscillator Hi All -
OK on the subharmonic thing. My question was the result of my mind locking
5502 2010-12-01 07:17:27 davidpnewkirk Re: Getting useful harmonic output from a BJT Colpitts oscillator